UWB antenna: control the path that carries timing

What makes an antenna suitable for UWB?

A UWB antenna must support the bandwidth, pattern and timing behavior required by the actual ultra-wideband radio configuration. In a ranging product, good matching alone is not enough. The signal path through the feed, antenna and installation can affect pulse transmission and measured timing. Select the channels first, then evaluate frequency response, installed orientations and the calibration method used by the system.

UWB is not one universal antenna footprint or one frequency span. A radio may support a limited set of channels within a broader UWB allocation. The antenna specification should follow those channels and the intended market, with the applicable limits confirmed through the product compliance process. A broad advertised bandwidth should not be treated as proof of positioning accuracy.

This guide focuses on antenna integration and measurement for UWB hardware. For general radiator selection and fabrication, use the antenna PCB hub. For board material and controlled RF construction, use RF PCB manufacturing. Neither route replaces the transceiver vendor's hardware and calibration guidance.

Begin with the radio and the timing task

Write down the radio part, supported channels, data mode and ranging architecture. Two-way ranging and synchronized multi-node systems place different demands on the full implementation. A single antenna path and a multi-antenna angular-measurement arrangement also need different physical and calibration information. Define the system before choosing an attractive radiator image.

The Qorvo DW3110 product documentation identifies channels 5 and 9 for that device and links hardware, antenna-selection, calibration and production-test resources. This is a concrete example of why the device configuration should own the channel requirement. It does not establish that every UWB product uses the same channels or that the manufacturer's advertised device performance transfers directly to an arbitrary PCB.

Use the vendor's current hardware guide for the selected device generation. A calibration procedure written for an earlier transceiver may be useful context but should not be assumed compatible without confirmation. Keep document revisions with the release so a future engineer can reproduce the basis of the design.

Make the feed, launch and radiator one reviewed structure

UWB feed transitions deserve attention across the required bandwidth. Pads, connectors, ground vias and changes of layer can introduce behavior that a uniform line equation does not capture. Preserve a continuous intended return path and avoid unreviewed stubs. Apply the antenna's own ground and keepout requirements at the radiator rather than removing copper indiscriminately from the complete RF route.

The selected laminate and construction affect electrical behavior, but a low-loss material name is not a substitute for a controlled geometry. State dielectric height, copper profile, finished copper, solder mask and relevant tolerances. Use high-frequency PCB material selection when comparing material families, and keep matched-condition data with that comparison.

An impedance calculation can support the feed-line design. The connector launch, antenna input and wideband response still require appropriate modeling or measurement. If a VNA connector is included only for development, define how the production path differs and whether that change needs another characterization.

Read phase as well as matching

Return loss describes reflection at a port. Transmission phase describes how phase changes through the measured path. Group delay is related to the slope of that phase versus frequency. A path can have an acceptable magnitude response while its phase slope varies enough to matter to a wideband waveform. Looking at both makes the engineering discussion more complete.

Keysight's group-delay measurement explanation describes the finite-frequency-aperture approach used by a network analyzer. The choice of aperture and sample density affects the result. A simple derivative can expose useful trends, but it can also amplify noise or conceal features when the frequency spacing is unsuitable.

For an over-the-air two-antenna measurement, the recorded transmission includes both antennas, their feeds, cables, free-space propagation and the fixture environment. It is not the delay of one antenna in isolation. Save separation, orientation, reference planes and any de-embedding method with the data. Changing the cable routing or antenna alignment can change the measured path even when the board is unchanged.

Convert unwrapped phase into interval group delay

Paste frequency in GHz and already-unwrapped S21 phase in degrees, with one sample per CSV row and no header. The worksheet computes group delay for each adjacent interval, the sampled minimum and maximum, ripple, span-weighted mean and largest frequency spacing. It keeps negative values visible for diagnosis and requires explicit confirmation that phase has already been unwrapped.

The calculation is minus the phase difference divided by 360 times the frequency difference. With GHz input, the result is in nanoseconds. A −450 degree phase change over 0.25 GHz gives 5 ns. The illustrative dataset introduces small interval changes around that value to show a 0.2 ns sampled ripple. It is a synthetic teaching example, not a measured UWB antenna or a ranging-accuracy specification.

The worksheet does not automatically unwrap phase. Sparse wrapped data can be ambiguous, and a plausible-looking derivative can be wrong if a cycle was missed. Obtain properly processed phase from the instrument or measurement workflow, then inspect sampling and reference-plane consistency. Increasing the number of points may be necessary before a comparison is useful.

Do not multiply this pair-path result by the speed of light and label it positioning error. A real ranging system includes its own timing definitions, calibration and signal processing. Group-delay variation, angular behavior and multipath need interpretation within that system. The output is intended to help an RF engineer identify questions and compare a defined measurement, not replace the radio's calibration procedure.

Keep calibration tied to the product revision

Antenna-delay calibration belongs to the complete device path and the vendor-defined procedure. Record the radio configuration, board revision, antenna and feed, housing, firmware, calibration setup and stored values. If a change affects that path, review whether calibration or characterization must be repeated. A value copied from an evaluation board is not automatically valid for a custom product.

Use a repeatable line-of-sight baseline before studying more difficult environments. Confirm the geometry of the test and the system's reference points. An enclosure dimension or a convenient tape-measure position may not match the physical or electrical reference assumed by the ranging method. Document the convention so later tests do not confuse an offset in the setup with a hardware change.

A scalar correction cannot repair every effect. Antenna pattern, orientation, polarization, nearby body or metal and non-line-of-sight propagation can create behavior that varies across use conditions. Characterize the application's realistic orientations and environment separately from the calibration baseline. Passing one straight-line distance test does not establish performance throughout a warehouse, room or moving installation.

Design around the installed tag or anchor

A compact mobile tag may sit near a battery and a person, while an anchor may be mounted against a wall, beam or metal bracket. These conditions can alter the antenna environment and the paths available for ranging. Include mounting details in the initial design review. A board that is excellent in free space may be unsuitable when the product's normal position blocks or detunes its useful direction.

For a device with multiple radios, evaluate simultaneous and representative operating states. UWB activity may coexist with Bluetooth discovery, processor loads and switching power supplies. Keep separate evidence for each RF function and then characterize coexistence. The BLE antenna integration page addresses installed packet testing; the IoT PCB page connects radio behavior to power and firmware responsibilities.

If multiple antenna paths are used for angular information, geometry and calibration become more demanding. Record their relative placement, feed implementation and any switching network. Do not assume that identical copper artwork guarantees identical electrical paths after assembly. The required evidence should come from the selected radio architecture and system measurement method.

Build an evidence set for the claimed application

Define test conditions before selecting a pass threshold. Use more than one sample when repeatability is important and retain the sample identity. Record temperature and operating state where they are relevant to the requirement. If results differ, inspect measurement repeatability before modifying a tuned antenna based on an unexplained single run.

The evidence should support the claim actually made to a customer. A useful product statement identifies the application and tested conditions. Avoid attaching a general centimeter-level promise to an antenna board because a transceiver product page advertises a capability under its own conditions. Board fabrication quality is important, but it is only one part of the positioning system.

Scope the PCB, assembly and RF verification separately

For a quotation, send the board outline, stackup, antenna geometry, radio reference, operating channels, BOM and quantity. Include the enclosure and mounting concept if integration is part of the discussion. State whether you need bare boards, populated hardware, programming, matching, calibration or a defined RF test. Do not assume these are all included in a basic PCB unit price.

The budget brief separates entered hardware, setup and verification allowances. No current UWB component prices, calibration time or laboratory fees are built in. A positive link or timing result does not populate a price automatically. If your design is still exploratory, leave the costs blank and describe the uncertainty; the PCB manufacturing route can then be discussed around the actual release and validation scope.

UWB antenna questions